About Agent Desktop
Agent Desktop is an MCP server published by BaLaurent in the Developer Tools category: an highly configurable desktop client for your favourite LLM.
The repository has 54 stars and 5 forks, with the last commit a month ago. Recent commit activity is the strongest signal that reported issues will get attention.
Install
git clone https://github.com/BaLaurent/agent-desktopThis server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.
Ask AI
Ask AI about Agent Desktop
Powered by Claude · Grounded in docs
Security checks
Static checks derived from the repository 6 hours ago. Heuristics on the source, not an audit.
- ✓README presentThe repository ships documentation.
- ✓License declaredUsable under a stated license.
- ✓Tests presentA test suite exists in the repository.
- !Dependencies pinnedNo lockfile. Installs can pull different code over time.
- !No dynamic code executionSource scan found eval or shell execution. Review how tool inputs reach it before granting credentials.
- !Scoped permissionsDoesn't declare a permission scope. Assume it can do anything its process can.
Releases
README
Agent Desktop
A desktop client for every LLM, built for agents. Claude by default — any model via a custom base URL. Plan mode approvals, MCP servers, a real git panel, a scheduler with variables, voice, Jupyter — wired straight to your filesystem, or headless in a Docker sandbox if that's your vibe.
Full documentation lives on the Agent Desktop Wiki — features, headless server, architecture, changelog, and guides.
Community
I build this project primarily for my own use — the features I add are the ones I need day to day. Feature requests are welcome on Discord.
Highlights
- Chat — streaming, markdown, syntax highlighting, message queue, conversation fork,
/compact,/clear - Auth — OAuth via Claude subscription or your own Anthropic API key
- Backends — Claude Agent SDK (default) or PI Coding Agent (experimental)
- Files — built-in explorer, Monaco code viewer, Mermaid, OpenSCAD 3D, Jupyter notebooks
- Knowledge base — attach files/folders as conversation context
- Quick Chat — global overlay with text + voice (local whisper.cpp)
- TTS — Piper, edge-tts, or spd-say with summary modes
- Scheduled tasks — recurring agent runs with prompt variables
- MCP — stdio, HTTP, and SSE Model Context Protocol servers
- Themes — full CSS custom property editor, day/night auto-switch, folder color tinting
- Git panel — status, branches, stash, DAG graph, log
- Headless mode — runs as a web server and/or Discord bot without Electron
- Auto-update — built-in, via electron-updater + GitHub Releases
See the wiki for the full feature list and configuration guides.
Installation
Prerequisites
- OAuth: an active Claude subscription + the Claude Code CLI (
npm install -g @anthropic-ai/claude-code), then runclaude login - API key: an Anthropic API key, configured in Settings > AI
Linux
Download from the Releases page.
# AppImage
chmod +x Agent-Desktop-*.AppImage
./Agent-Desktop-*.AppImage
# Debian / Ubuntu
sudo dpkg -i agent-desktop_*.deb
macOS (Apple Silicon)
Download the .dmg from Releases, drag to Applications, then right-click → Open on first launch (unsigned app).
Windows
Download from Releases — NSIS installer or portable .exe.
Development
git clone https://github.com/BaLaurent/agent-desktop.git
cd agent-desktop
npm install
npm run dev
npm run build # compile TypeScript (output: out/)
npm run dist:linux # AppImage + deb
npm run dist:mac # .dmg (arm64)
npm run dist:win # NSIS + portable
npm test # main + renderer
Architecture, IPC patterns, theming, headless deployment, and contribution guides → wiki.
License
README fetched from the repository 6 hours ago. The original is authoritative.